From: Chris Hibbert <hibbert@netcom.com>
Date: Mon, 5 Sep 94 18:05:12 PDT
To: cwalton@earthlink.net (Conrad Walton)
Subject: Re: Reputations/Credentials
In-Reply-To: <m0qhk6b-000LFEC@moon.earthlink.net>
Message-ID: <199409060105.SAA08869@netcom6.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain



I'm not going to quote Conrad Walton point-by-point, but I'm
responding to the message in which he asks how what TRW provides
relate to what we mean by reputations and credentials.


What TRW does is to collect information from others about their
beliefs about others and their history of dealings with them.  TRW
then provides a summary giving their opinion.  They do it in an
automated way, and provide a numerical rating as the output.  Equifax,
TransUnion and Dunn&Bradstreet provide a similar service, but depend
on different sources, and combine the information in different ways.
I doubt if any of them would tell you what their formula is.

I think what other c'punks writing on this topic have objected to is
the notion that someone might create *a* calculus that would describe
*the* proper way for rating services to do their job.  Reputations are
people's opinions, and how you add them up depends on your beliefs
about the opinion-holders.

I'm not sure that credentials are different in that respect.  The way
credentials should be different is that they should tell you what
opinion they're intended to represent.  Does your signature on my key
indicate that you believe that I'm a real person with the name I use,
or just that I am the person who used that name last year?

Reputations are subjective.  Credentials are codifications about
beliefs.  They say that X believes Y about Z.  It might be useful to
codify what the different useful Y's are, but I find it hard to see
how there could be a general formalism for composing statements like
these.

Chris